Please help, parametric or non-parametric test?

I have a data set of 58 records containing students, gender, total sat score and sat score for each part. 40% of the records are Male, 60% female

Question 1:

i want to compare means of male/female for EACH SAT part, is the sample large enough to assume normal distribution and use a t test or should i use a non-parametric test? I know the Wilcoxon Rank Sum is a cousin of the t-test; however, I believe one of the assumptions is that the distributions are the same and when I did histograms for each parts, for some the distributions were not the same.

Or should I use one test for those that appear to have the same distribution and different one for those that have different distributions?

Question 2:
I'd like to know the answer to this regardless of whether I use Wilcoxon or not, I'd like to know the answer to this -- how would I state my hypothesis if using the rank sum?